A Shipping Crate for the v3.0 Attenuator and …

For a little while I have wanted to experiment with laser cut plywood boxes. There are software tools available to design the box joints at the corners and I wanted to try these. A 1/8” thick plywood “shipping crate” for the v3.0 attenuator provided the perfect excuse. The pictures below show the completed crate.

v3.0 Attenuator “Shipping Crate” closed.

v3.0 “Shipping Crate” open.

The shipping crate measures 22 cm x 8.5 cm x 9 cm (L x W x H) and sports a black plastic draw latch to keep it closed. I’m very happy with the results.

DÆ phono preamp v2.1 Update

I have also continued testing various design iterations of the v2.1 Phono Preamp and have achieved the following excellent results:

• RIAA accuracy of +/- 0.05 dB from 20 Hz to 20 kHz;

• Noise less than 35 uV with a gain of 40 dB at 1 kHz and 100 Ω source resistance;

• Distortion of 0.0025% at 1 kHz and -45 dBV input. The distortion was almost equally as low from 100 Hz (and below) to over 5 kHz where it started to rise. The distortion at 1 kHz was less than 0.009% and 1.7 Vrms output.

I am in the process of designing a new printed circuit board based on the latest tests and will have this ready for listening tests by mid February.

New Complete Preamp

Also in the near future I will redesign the complete preamp enclosure to house the v3.0 attenuators and v2.1 phono preamp. More to come on this…

Phono Preamp v2.1

As promised, here is an update on the latest DÆ phono preamp design. After some listening experience and testing with the QuantAsylum QA401 Audio Analyzer I made a few changes to my phono preamp design. The test results for the v2.0 phono preamp showed very low noise and a very accurate RIAA response curve as expected. The circuit uses a gyrator to simulate an inductor to establish the DC operating point. In the literature, the gyrator is normally shown connected to ground and simulates a ground connected inductor. I found that connecting the gyrator to a DC supply voltage other than ground also worked but - the gyrator circuit is implemented with op-amps that run very hot even after a few are used in parallel to share the DC bias current. Also the v2.0 phono preamp had higher distortion than I wanted and I attributed this to the feedback assisted current mirrors and gyrator circuits that I used.

Phono Preamp v2.1

A pdf of the schematic for the v2.1 phono preamp is in the link. Click the word “schematic” to retrieve the pdf. The v2.1 design uses emitter degenerated BJT current mirrors to set the DC conditions and also mirror to signal current from the input differential pair over to the RIAA filter. This design is much less expensive but - there are still a few issues I need to solve. The distortion is actually greater at low frequencies than the v2.0 design and I may go back to a feedback-assisted current mirror for at least the signal current. However I will use an opamp that can better handle the high common mode voltage than the venerable NE5532.

Also the BJT current mirror transistors used to set the bias conditions run hot because there is 20 mA running through them and a voltage drop of 14 volts or 280 mW of dissipation. The transistors can handle this but any air current passing the transistors changes the bias conditions slightly. The changes in bias are quickly compensated by the DC servo but I would rather mount the transistors on a heat sink to reduce the influence of air currents. The same applies to the transistors used in the input differential pair.

Finally the v2.1 design has some excess attenuation at 20 Hz compared to the v2.0 which is flat down to 20 Hz. This due to the changes I made to the DC servo design and I may revert back to the v2.0 design.

More to come…

The v3.0 Attenuators are coming

It has been almost three months since my last blog post. I have been busy tending to other things but I have also been busy advancing the v3.0 attenuator and the v2.1 phono preamp (more on this later). The photo below shows the rear of the right channel v3.0 attenuator. I have so far built one complete pair for testing.

Rear of v3.0 attenuator.

The v3.0 attenuator is a radical departure from the v2.0 series attenuators that used a vacuum chamber to reduce relay switching noise. Instead of a vacuum chamber, the v3.0 attenuators use tiny reed switches and a few reed relays to perform the switching duties. The use of reed switches and relays allows the v3.0 to be significantly smaller and less costly than the v2.0 series attenuators.

The picture below shows the ladder section of the v3.0 attenuator. There are two circular attenuator levels shown with 12 steps on each level. This attenuator has a total of 24 steps. The v3.0 design allows for a total of eight levels or 96 total levels if desired! The level selection is preformed using tiny magnets on a “yoke” (red part in picture) that is rotated into position with a stepper motor (silver parts just to the left of center in the picture) to ensure accurate level selection.

Ladder section of v3.0 attenuator.

The v3.0 design includes two stepper motors; one stepper motor (black part at left side of picture) for the knob and the second motor to perform the level selection. Each of the two stepper motors are controlled by a Trinamic driver in StealthChop mode to ensure silent operation.

The testing of the v3.0 attenuator is ongoing but so far the results have been spectacular!

QuantAsylum QA401 Audio Analyzer

One of the tasks I have on my to do list is to increase the gain of my phono preamp v2.0 from 35 dB to 40 dB at 1 kHz. Before I update the design I wanted to benchmark the performance of the existing design. I read about an audio analyzer called the QuantAsylum QA401 Audio Analyzer in book by Jason Stoddard about the start-up of Schitt Audio entitled “Schiit Happened: The Story of the World's Most Improbable Start-Up” (an excellent book by the way). QuantAsylum QA401 Audio Analyzer is capable of measuring the noise, frequency response and distortion of the phono preamp.

QuantAsylum QA401 Audio Analyzer

So far I have been able to confirm that the preamp implements the RIAA equalization curve to +/- 0.025 dB from 20 Hz to 20 kHz. This is exceptionally accurate and is the result of using capacitor multipliers and fine tuning as described in my November 30th, 2018 post. The noise level is also very low with an output noise level of 18 uV with 100 Ω input source resistance (-94.9 dBV or - 92.7 dBu) again from 20 Hz to 20 kHz. The Spice simulation of the circuit reported 10 uV of output noise so I am still trying to track done any extraneous noise before recording a final measurement. I still need to complete the distortion measurement (THD).

I will post graphs of all this data once I complete the testing.

This is Glue. Strong stuff.

Sorry for the Blues Brothers quote - I couldn't help myself.

Not these Blues Brothers.

Photo by jpenner/iStock / Getty Images

I couldn’t afford to license a Blues Brothers image, but if you are curious check out this YouTube video. And then if you haven’t seen it check out the whole movie.

For the past eight months I have been working on a cable feed through that uses Torr Seal epoxy. Torr Seal is a specifically formulated low vapor pressure epoxy that is meant to seal vacuum leaks at very low pressures. A key feature is it’s very low off-gassing. It is a great product but I am now wondering if it also has some downsides for my application. Torr Seal is a thick product with the data sheet listing the viscosity as “Thick non-flow paste”. Also it appears to be a very rigid epoxy although its elongation (amount it can stretch when cured) is not listed on the data sheet. Finally the data sheet says “Will not adhere to Teflon, Kel-F, nylon nor polypropylene”. While not specifically listed I wonder how well it adheres to acrylic plastic. Torr Seal may work well with the aluminum end plates at least most of the time but a source of leaks for the all acrylic enclosure. The vacuum preparation of the Torr Seal appears to improve the situation but it is still not satisfactory especially for the acrylic version of the enclosure.

The upshot of all this second guessing is I have started a test program for a few different adhesives/sealants. Here is a list of the products I will try:

Product	Formulation	Viscosity [cP]	Elongation [%]
Torr Seal	Epoxy	Thick non-flow paste	Not in data sheet
1C-LV	Epoxy	66,750	3
Plastic Welder II	Acrylic	55,000	5
H3000	Acrylic	45,000	20
E-30CL	Epoxy	10,000	8
U-09FL	Urethane	5,000	175

I plan to test each of these products on a 9 mm acrylic end plate with and without vacuum preparation to remove air pockets. I started with the 1C-LV epoxy and right away the benefits of the lower viscosity are apparent - the 1C-LV readily filled the cable feed through opening in the acrylic with no air pockets even without vacuum preparation. A key feature of three of the products (Plastic Welder II, E-30CL and U-09FL) is the data sheets specifically state they are for use on acrylic plastic.

I have started testing the 1C-LV and results look promising.

Finally a Sealed v2.2 Acrylic Enclosure

After many trials and tribulations I have finally managed to assemble a vacuum tight acrylic version of the v2.2 attenuator enclosure. The two advances tested with this prototype are the 9 mm acrylic end plates and the vacuum preparation of the Torr Seal epoxy cable through to remove air pockets.

Increasing the thickness of the acrylic end plate from 5.6 mm to 9 mm has made the end plate a great deal more rigid. This improves the compression of the o-ring and eliminates leaks at the o-ring between the corner fasteners.

For sure using a vacuum to degas the Torr Seal epoxy in the cable feed through opening in the acrylic was a big improvement. Looking through the acrylic end plate you can see that the epoxy now fills the entire cable feed through opening. There are no more air pockets in the epoxy that might lead to a leak. There may be tiny air bubbles in the epoxy but the air pockets that could join to make a leak are gone. The only thing I don’t like about the vacuum prepared epoxy feed through is that the epoxy actually expands during the preparation and now overflows the opening in the acrylic end plate. The overflow probably just improves the seal but I hate the look of it. I need some more finesse in this part of the design/assembly process.

Both improvements look very promising and the vacuum has been maintained for 24 hours. I will keep watching this for a week or more as I work on other things. Next up I will finish repairing the v2.2 aluminum/glass version by making a new rear end plate and use the vacuum prepared Torr Seal technique. I will also order parts to make a complete acrylic version.

Acrylic version of v2.2 holding a vacuum.

More Vacuum Leak Troubleshooting

As I mentioned in my last post, I had built four v2.2 attenuators and three of them were free of leaks and have held a vacuum for weeks now but one of the attenuators had a slow but noticeable leak. I have a few tests planned to pinpoint the source of the leak. So far I know that the leak appears to be in the rear aluminum cap of the vacuum chamber. The leak could be in one of three places: 1) the Torr Seal epoxy cable feed through; 2) the threaded hole for the sleeve valve; 3) the Viton seal between the aluminum cap and the glass cylinder. I plan to evaluate each of these potential sources for the leak.

At the same time I am building a version of the vacuum chamber with 9 mm acrylic end plates. I have one acrylic end plate ready to test including the cable feed through with Torr Seal Epoxy. One feature (advantage/disadvantage?) of the acrylic end plate is that I can see the quality of the Torr Seal epoxy fill in the cable through because the acrylic is clear. The conclusion is that while the surface of the epoxy appears to completely fill the cable feed through opening in the acrylic, there are air pockets in the epoxy that are clearly visible; so the cable feed through may not be as solid as I thought previously. It may be that the epoxy fill is a bit hit and miss and the air pockets could be a weak point where a leak could develop.

To tackle this problem I have bought a vacuum degassing chamber so that I can put the Torr Seal epoxy joint under vacuum before the epoxy cures. I call this vacuum preparation of the epoxy. I expect that this will avoid the air pockets in the epoxy cable feed through and reduce the chances of a leak at this point.

Vacuum degassing chamber.

And then there were 2…

I have finally successfully assembled two v2.2 headless attenuators and they work beautifully. Electrically they work exactly as designed and the vacuum has held steady for over a month.

v2.2 Headless attenuators!

There is work to do on this yet - I want to redesign the wooden base and the power supply box (central black box). The base and power supply box were designed for the v2.1 version of the attenuator that had a vacuum sensor in the power supply box along with vacuum hose between the power supply box and the attenuators. There was also a vacuum hose emanating from the rear of the power supply box to a ball valve. All of the external vacuum hoses have been eliminated in the v2.2 design by using sleeve valves and a vacuum sensor inside each vacuum chamber. This has greatly simplified and “cleaned-up” the design. A big improvement and one of many improvements between the v2.1 and v2.2 designs.

v2.2 Headless attenuators and associated remote.

v2.2 Headless attenuators rear.

Actually there are three attenuators because I also assembled two complete v2.2 attenuators with the motorized knobs but one of them has a slow vacuum leak that I am trying to troubleshoot. I will post pictures of the two complete v2.2 attenuators as soon as I solve the vacuum leak (soon I hope).

v3.0 Attenuator

The readers will be happy to know that the headless v2.2 attenautor I wrote about in my last post on June 27th has now held a vacuum for two weeks without any measurable air infiltration!

While I wait for the last few parts for the v2.2 attenuators, I have been designing and collecting parts for the v3.0 attenuator. As I described in the May 9th post, the v3.0 attenuator is based on reed relays instead of telecom relays in a vacuum chamber. I am still in the design/testing phase for the v3.0 attenuator but here are a few pictures to show the progress.

v3.0 attenuator front.

v3.0 attenuator top.

In the top view there is a red “yoke” that holds two small magnets. The yoke is attached to a small stepper motor (grey motor just to the left of centre in the photo) that will rotate the magnets in front of a series of reed relays and associated attenuator resistors. The reed relays and resistors will be soldered to the edge of the PCBs in the silver “Half-cut/Castellated Holes”.

The larger stepper motor (black motor just to the right of centre in the photo) supports the input knob shaft. A larger motor is desired here to provide an improved tactile feel while the owner turns the knob manually. The improved feel is provided by magnetic damping caused by electrically shorting the stepper motor coils. If the input shaft stepper motor is too small, it doesn’t have enough torque to provide the proper feel.

v3.0 attenuator rear view.

The photo of the rear view shows two layers for attenuator resistors. Each layer will have 12 steps. A total of eight layers or 96 steps will be possible. I still need to design and build the rear cover that will include the RCA connectors and a cover for the attenuator sections.

v2.2 attenuator (left) and v3.0 attenuator (right) compared.

The photo above shows the size difference between the v2.2 attenuator (left) and v3.0 attenuator (right). The v3.0 is considerably smaller than the v2.2 attenuator. For comparison the front face of the v2.2 attenuator is 3 5/8” square whereas the front face of the v3.0 attenuator is 2 3/4” square or about 25% smaller. The overall length of the v2.2 attenuator is 7 1/8” whereas the v3.0 is expected to be 5 1/4” long are about 25% shorter.

Eureka ... at long last

I noticed that it has been just over a month since my last post but this is a big one.

After many months of struggling to make a prototype on my v2.2 attenuator, I have finally successfully assembled one. It is like the Thomas A. Edison quote, “I have not failed. I've just found 10,000 ways that won't work.” It is more like 20 ways in the case of the v2.2 attenuator but it feels like 10,000.

Pictured below is a headless version of the attenuator. Headless because it does not have a knob or switches to make adjustments but instead it is controlled by a remote. I have completed the initial electrical tests and it works as designed! It has also held a vacuum for over three days now without any noticeable air infiltration. It is remarkably quiet when adjusting the gain or changing the selected input.

Next up - finish assembling the left channel mate for the attenuator pictured below and then complete a pair with the motorized knobs and input selector switches. I have over 90% of the required parts so after I order the last remaining parts, it is a matter of some really careful assembly work.

Front right corner of v2.2 headless attenuator.

Top of v2.2 headless attenuator.

Rear left corner of v2.2 headless attenuator.

Troubles

Again one step forward and two steps back (well maybe only one step back).

As described in my April 29 post, I changed the flex cables from 1 mm pitch to far more flexible 0.5 mm pitch cables. This change was a big improvement and the ladder attenuator cube is now resiliently mounted. Even without the vacuum the attenuator is very quiet because there is no mechanical transmission from the relays to the outside world.

I thought I was home free - well not so fast.

0.5 mm flex cables at both ends of the attenuator.

To install the new cables I had to desolder a small PCB from the vacuum cable feed through. I call this PCB “the pitch converter” because it allows the connection of a 0.5 mm pitch flex cable to the 0.1 inch pitch vacuum feed through. There are 12 connections and it is necessary to heat all 12 at one time to remove the PCB. Well it appears that applying enough heat, long enough to remove the PCB damages the Torr Seal epoxy surrounding the cable feed through pins. I confirmed this by trying one of vacuum chambers before removing the pitch converter PCB (it held vacuum as designed) and after removing the pitch converter PCB (it leaked).

“Pitch Converter” PCB.

The only way to fix this is to make new aluminum end plates and associated components. I now know what I will be working on next month…

This news is painful to the pocket book and more importantly another loss of momentum. I am really surprised how finicky and unforgiving this design is and flawless execution is required at every stage. There appears to be very few ways to correct any assembly mistakes without major re-work.

On the bright side I have tested all the design features of v2.2 at least twice and when I finally manage to assembly it without errors or last minute design changes the final product will be excellent.

Into the Reeds

As I wait to receive the next batch of PCBs for the v2.2 attenuator I started thinking about how to further improve my design. I went back to consider my objectives which are:

• Make the attenuator feel like a high quality potentiometer volume control. I didn’t want a stepped attenuator based on a rotary switch with detents. The stepper motor I use as an input device satisfies this objective;

• To build a stepped attenuator for better accuracy than what is available with a conventional potentiometer based volume control. A potentiometer uses two resistor traces for the left and right channel. A stepped attenuator uses a string of resistors and is known to have much better left/right channel balance and can follow a desired taper (attenuation vs rotary position) with much greater accuracy. Stepped attenuators often have one resistor for each ladder step. I decided to use two parallel resistors for each ladder step to further enhance accuracy so I think this objective is achieved;

• The volume control should be motorized so that the knob would actually rotate in response to commands from a remote control. I like the aesthetics of this;

• I wanted two totally separate attenuators for the left and right channel so that either could be rotated and the other attenuator would follow. I think this looks cool and the feature has been realized by a Bluetooth link to coordinate the two attenuators;

• Make it silent, meaning no electrical glitches/switching noise, no audible clicks from the relays and also no noise from the motor operation. The Trinamic motor driver with StealthChop mode has made the stepper motor movement silent - so ✔ on that objective. The software makes sure that at least one relay is energized at all times in a make-before-break feature so the electrical glitches/switching are eliminated when switching between steps - so ✔ on that objective. And of course the vacuum chamber is there to reduce audible clicking noise - almost ✔ on that objective too - just need to pull it all together.

It seems I achieved my objectives.

None-the-less the world is not perfect yet and further improvements are possible.

One improvement might be to reduce the size so I started to search the web for smaller relays because the relays along with the ladder resistors take up the most board space. I think I am already using one of the smallest electromechanical relays available at 10.6 mm x 7.4 mm x 10 mm.

During the search I found some very small reed relays at 3.9 mm x 3.9 mm x 15.5 mm which would certainly be interesting. The reed relays also have the benefit of being virtually silent without the vacuum chamber. The downside appears to be cost because a reed relay is about three times the price of the electromechanical relays. The total cost of relays adds up because my current design uses one relay per ladder step. I will look at this further at another time but I thought of a way to use a simple reed switch and a rotating magnetic instead of a reed relay. The reed switch is much less expensive than a reed relay because there is no coil or associated enclosure.

It turns out this is not a unique idea (nothing is really new under the sun) but it may be possible to implement this idea using some of the techniques I learnt for the vacuum stepped attenuator to design a new version of this idea.

v3.0 of the DÆ stepped attenuator of the on the horizon?

Reeds!

Flex Cables

In yesterdays post, I mentioned the need to update the flex cable design. The flex cables connect the vacuum tight cable feed through to the ladder PCBs. The idea is to have a resilient mount between the attenuator ladder sections and the enclosure and the flex cable is part of the resilient mounting system. The resilient mount breaks the mechanical transmission from the relays to the outside world and the vacuum breaks the air-borne transmission path.

While assembling the v2.2 version of the atttenuator, I found two problems with the flex cables: 1) the flex cables are too long which pinned the ladder attenuator PCBs to the top of the vacuum enclosure; 2) the flex cables are too stiff. To correct these two problems I are going to use a narrower, more flexible cables and lower the mounting point to make better use of the available flex cable length. Two inches is a standard length flex cable length and the next shorter length is a bit longer than one inch which is far too short. The current cables are 1 mm pitch flex cables and I am going to modify the design to use 0.5 mm pitch flex cable.

Ladder sections with white flex cable.

The picture below compares the 1 mm pitch cable I have been using to the 0.5 mm pitch cable I am going to use. Both cables have 12 conductors and are two inches long. The 0.5 mm pitch is significantly more flexible that the 1 mm pitch cable. I need to make a minor revision to three PCBs to accommodate the change to the narrower cable. I should complete the design and order the new PCBs this week and be back at it in a couple of weeks when the new boards arrive.

Comparison between 1 mm pitch (left) and 0.5 mm pitch (right) flex cables. Note that the 0.5 mm cable has a connector attached at the top.

Ladder Sections Repaired

Over the past week, I repaired the ladder sections for the right channel. I had installed the ladder resistors in the wrong order. One of the ladder sections is shown in the picture below. The lowest ladder rung (least gain) is at the upper left and the the rungs increase in a counter-clockwise direction around the perimeter of the board. The resistors are the blue parts with the colored bands. I had mistakenly installed the resistors starting at the upper right and in a clockwise direction which resulted in a weird gain curve that looked nothing like the audio taper I was looking for. To correct this, I carefully desoldered the first set of resistors and soldered a new set of resistors onto the boards. I almost completed the right channel but in the process damaged the PCB traces on one of the ladder sections and need to make one new section. Fortunately I bough some spare PCBs.

With the corrected resistor arrangement, I did some initial testing of the attenuation curve and it is looking good. Final testing still to come.

Repaired ladder section.

While looking at a close-up of a ladder section, I want to point out a few things. For scale, the PCB is two inches (50 mm) square. Each ladder section has eight rungs (or settings or steps). There are eight tiny relays which are the ivory colored parts in the picture with the word “Japan” on them. Next to each relay are two parallel resistors which are the blue parts with colored bands. The resistors are precision 1% metal film resistors and two resistors are placed in parallel for improved accuracy. Often ladder attenuators will only have one resistor for each rung so doubling the number of resistors is a feature of my design.

The ladder sections are stacked to make a ladder attenuator with the desired number of steps. The picture below shows a 24-step attenuator using three ladder sections. At the front of the ladder stack is a connector board with the flex cable attached. The box like component next to the flex cable is a vacuum sensor that will allow the vacuum level in the chamber to be monitored. The fifth board, at the rear of the stack, has four relays used to select one of four input sources namely Phono, CD, Aux 1 and Aux 2.

The attenuator stack is suspended in the vacuum chamber by the metal rod and four small rubber bands. The mounting is meant to be resilient to prevent mechanical transmission of vibration from the relays to the enclosure. The vacuum stops the air-borne transmission of relay switching noise and the resilient mount prevents the mechanical transmission of switching noise.

Good idea but only problem is the flex cables are too long and stiff - I am working on correcting this and will talk about it in a future post (maybe tomorrow?).

A 24 step attenuator.

v2.2 Assembled

Well here it is - an assembled v2.2 Stepped Attenuator. The assembly went fairly smoothly; actually very smoothly compared to previous versions. Some of the improvements I made where aimed at improving the ease of assembly. I started the electrical testing and have run into a few minor snags like finding out that I installed the ladder resistors in the wrong order. Oops. This is nothing that a little careful desoldering/resoldering can’t fix. I also have a problem with the length of the flex cables that are part of the resilient mount design for the relay ladder. The flex cables are too long pinning the relay ladder against the top of the vacuum chamber. I have ordered shorter flex cables which should arrive in a couple of days. I may also try an even smaller pitch version of the flex cables if the shorter cables are too rigid but this requires updating three of the printed circuit boards (again!).

Within a week or so I should have all the electrical testing complete and the minor problems sorted out. At that point I will evacuate the vacuum chamber and have completed the v2.2 attenuator. Actually two v2.2 attenuators for a stereo pair.

Next, I have updated the remote design (I already have the parts for this) and have plans to update the phono pre-amp to get some more gain. Oh ya, I also have also all the parts for a pair of “headless” v2.2 attenuators used in the passive pre-amp design. Lots to do…

v2.2 Stepped attenuator front corner.

v2.2 Stepped attenuator side view.

v2.2 Stepped attenuator rear corner.

Making Progress

I made good progress this week assembling and testing the Panel Power, Panel Source and Control boards for the left and right channels of the v2.2 stepped attenuator. Here are a few of the results:

Panel Power, the PCB for the front panel power switch:

• The light sensor is a big improvement. The light sensor is used to control the brightness of the power and source LEDs according o the room light level;

Panel Source, the PCB for the front panel source selector switches:

• LED brightness control tested and shown to function well - a big imrovement;

• The simplified schematic using standard logic gates instead of a programmable device (CPLD) has improved the speed of construction because the programming of this board is no longer required;

Control, the PCB with the Bluetooth radio module (with micro-controller) and stepper motor driver:

• Change ladder step selection logic to reduce connections to the ladder from 18 to 12 - this has not been tested yet;

• Modify the board to use the improved vacuum chamber cable feed through - again not tested yet;

• The Trinamic IC StealthChop mode provides almost silent operation of the stepper motor - it is wonderful to see (and not hear) the volume control knobs glide silently up and down the dial;

• Add light sensor so the brightness of the power LED is tailored to the room light level. This is only used for the “Headless” version as shown in the DÆ Passive Preamp v2.1 - I will build a pair of the headless version after I complete the version I am working on now.

I have started constructing the other boards and have already finished one of the Front Connector Plates (see description of Front Connector Plate in my March 18th Blog post). Next week I will finish all the boards for at least one ladder section. I had to make a few last minute changes to the Rear RCA Connector board and re-order these boards. The boards are now being manufactured. I hope to get these updated Rear RCA Connector boards in the third week of April so I am not delayed too long.

v2.2 Stepped attenuator under construction and testing.

Just Arrived

This past week, I received the printed circuit boards (PCBs) for the v2.2 Stepped Attenuator. The attenuator includes all the improvements listed in my March 18th post. There were 51 boards in the shipment; this is enough to build two motorized attenuators and two headless attenuators. My plan for the next few days (Ok weeks) is to populate the boards and test the complete attenuators in the glass/aluminum version of the vacuum chamber.

With any luck these will be the first fully functional versions of the vacuum based stepped attenuator.

Wish me luck…

Knock on wood…

Control board for the DÆ Stepped Attenuator v2.2.

v2.2 Stepped Attenuator

Last week I ordered the printed circuit boards (PCBs) for v2.2 of the stepped attenuator. There are numerous improvements in the v2.2 version compared to the v2.1 version. Some of the improvements include:

Panel Power, the PCB for the front panel power switch:

• Add light sensor so the brightness of the power and source LEDs are tailored to the room light level.

Panel Source, the PCB for the front panel source selector switches:

• Change schematic to permit LED brightness control;

• Simplify the schematic to use standard logic gates instead of a programmable device (CPLD).

Control, the PCB with the Bluetooth radio module (with micro-controller) and stepper motor driver:

• Change ladder step selection logic to reduce connections to the ladder from 18 to 12;

• Modify the board to use the improved vacuum chamber cable feed through;

• Change motor driver IC to Trinamic with called StealthChop mode;

• Add light sensor so the brightness of the power LED is tailored to the room light level. This is only used for the “Headless” version as shown in the DÆ Passive Preamp v2.1 ;

Front Connector Plate, the PCB used to connect the flex cable from the Control PCB to the ladder boards:

• Reduce the number of flex cable positions from 18 to 12;

• Add absolute pressure sensor so that the vacuum level can be monitored from the inside of the vacuum chamber. This reduces the potential for vacuum leaks by eliminating the number of external vacuum connections;

• Add voltage regulator for absolute pressure sensor for improved accuracy;

• Improved grounding and shielding to reduce noise;

• Round corners of the board to reduce chances of resiliently mounted boards touching the vacuum chamber walls;

• Added holes for retaining bolts to hold the Front Connector Plate, Ladder sections and Source PCBs together for enhanced reliability;

Ladder, the PCB that implements that actual ladder sections. There are two flavors called even and odd that used in an alternating pattern to implement the ladder:

• Reduce the number of board-to-board connector positions from 18 to 12;

• Simplify schematic slightly reducing the number of components;

• Improved grounding and shielding to reduce noise;

• Round corners of the board to reduce chances of resiliently mounted boards touching the vacuum chamber walls;

• Added holes for retaining bolts to hold the Front Connector Plate, Ladder sections and Source PCBs together for enhanced reliability;

Source, the PCB is used to select the source (PH = phone, CD, A1, A2):

• Reduce the number of board-to-board connector positions from 18 to 12;

• Simplify the schematic to use standard logic gates instead of a programmable device (CPLD);

• Simplify schematic slightly reducing the number of components;

• Improved grounding and shielding to reduce noise;

• Round corners of the board to reduce chances of resiliently mounted boards touching the vacuum chamber walls;

• Added holes for retaining bolts to hold the Front Connector Plate, Ladder sections and Source PCBs together for enhanced reliability;

Rear RCA Connector, the PCB that has the rear RCA connectors:

• Reduce the number of flex cable size from 18 to 12 positions;

• Modify the board to use the improved vacuum chamber cable feed through;

• Add ground pins to improve grounding by connecting the rear RCA connector PCB to the rear aluminum end plate of the aluminum/glass version of the vacuum chamber. This will also ground the front aluminum end plate because the fasteners electrically connect the front and rear aluminum end plates.

So only a minor update from v2.1 to v2.2.

Stepped attenuator v2.1. The overall resistance of the ladder is 10 kΩ and there are 24 steps. Note tiny white telecom relays and that two parallel resistors are used for each ladder section for improved accuracy.

Sorry for the Long Delay...

It is hard to believe it has been one month since my last post but I haven’t been slacking!

There is been some bad weather in Canada with snow and ice storms and, yes, that has slowed the testing and prototyping work down. In particular it takes and extra day or two for shipping (feels more like a week longer). I finally received the top and bottom covers for the remote and it came together nicely.

The remote completed!

I received more parts for the glass and acrylic vacuum enclosures. I found some really nice Brass Heat-Set Inserts for Plastic at McMaster-Carr that will allow me to use the sleeve valve on the acrylic enclosure. The inserts are heated with a soldering iron and while hot pressed into the acrylic. This worked quickly and very well but of course I will need to seal the joint with Torr Seal epoxy if there is any hope of it being vacuum tight.

I completed testing a very quiet stepper more driver IC from a German company called Trinamic. The IC has a mode called StealthChop which is advertised to make the stepper motor operation “completely noiseless”. It is truly quiet! I am now testing the last miscellaneous auxiliary circuits and will soon be completing the schematics and printed circuit boards for the v2.2 version of the stepped attenuator.

I am so looking forward to pulling the v2.2 version together.

More on this later…

Off to the Races

DÆ Remote

I received the printed circuit board for the remote and had time to populate it. I also had time to advance the firmware to a near final version. I received the laser cut acrylic top and bottom but I am still waiting for the 3D printed covers. I should have the covers by the end of the week and will add a better description to the DÆ Remote product page when I get that far.

I am very happy with the results. It looks so cool that my wife wants one even though she doesn’t need it. High praise in anybody’s book. Actually she asked if I can turn the industrial design into a phone. Another future project?

The remote uses capacitive touch technology so the main body is only 6 mm thick! Using proximity sensing it lights up when you go to pick it up. Turns out this is a fun feature in a brightly lit room but really works well in a dimly lit room; and a lot of serious listening in done in a dimly lit room. The remote has a light sensor so the intensity of the back light is tailored to the ambient light in the room. This seems like a small detail but I need to add it to my pre-amp design because the indicator lights on the pre-amp are good during the day but far too bright and distracting in lower ambient light. The lithium battery is charged by plugging the remote into a USB port. I expect the remote will run for a few months on a single charge.

The remote is coming together.

Vacuum Testing, Vacuum Testing and More Vacuum Testing

I think I have now solved all my major design issues with the glass vacuum chamber. The sleeve valve, improved cable feed through and absolute vacuum sensor all work well. The absolute vacuum sensor is on the PCB inside the vacuum chamber in the picture below. Using this sensor, I can draw a vacuum through the sleeve valve, close the sleeve vale and plug its port while still getting a vacuum measurement. This design eliminates the external vacuum tubing which is a potential source of leaks.

I have now tested for almost a month with very, very little pressure change. Maybe only the small amount of off-gassing from the components is contributing to the tiny decrease in vacuum I can measure. The absolute vacuum sensor saturates as all the air is evacuated and the leak rate is now approaching the limit of what I can measure. Perhaps it is time to add a “getter” material for the next bit of performance improvement.

Long-term vacuum testing including glass enclosure, sleeve valve and absolute vacuum sensor.